Initial commit - Dutchie dispensary scraper

backend/src/services/scheduler.ts

@@ -0,0 +1,116 @@
+import cron from 'node-cron';
+import { pool } from '../db/migrate';
+import { scrapeStore, scrapeCategory } from '../scraper-v2';
+
+let scheduledJobs: cron.ScheduledTask[] = [];
+
+async function getSettings(): Promise<{
+  scrapeIntervalHours: number;
+  scrapeSpecialsTime: string;
+}> {
+  const result = await pool.query(`
+    SELECT key, value FROM settings
+    WHERE key IN ('scrape_interval_hours', 'scrape_specials_time')
+  `);
+  
+  const settings: any = {};
+  result.rows.forEach(row => {
+    settings[row.key] = row.value;
+  });
+  
+  return {
+    scrapeIntervalHours: parseInt(settings.scrape_interval_hours || '4'),
+    scrapeSpecialsTime: settings.scrape_specials_time || '00:01'
+  };
+}
+
+async function scrapeAllStores(): Promise<void> {
+  console.log('🔄 Starting scheduled scrape for all stores...');
+  
+  const result = await pool.query(`
+    SELECT id, name FROM stores WHERE active = true AND scrape_enabled = true
+  `);
+  
+  for (const store of result.rows) {
+    try {
+      console.log(`Scraping store: ${store.name}`);
+      await scrapeStore(store.id);
+    } catch (error) {
+      console.error(`Failed to scrape store ${store.name}:`, error);
+    }
+  }
+  
+  console.log('✅ Scheduled scrape completed');
+}
+
+async function scrapeSpecials(): Promise<void> {
+  console.log('🌟 Starting scheduled specials scrape...');
+  
+  const result = await pool.query(`
+    SELECT s.id, s.name, c.id as category_id
+    FROM stores s
+    JOIN categories c ON c.store_id = s.id
+    WHERE s.active = true AND s.scrape_enabled = true
+      AND c.slug = 'specials' AND c.scrape_enabled = true
+  `);
+  
+  for (const row of result.rows) {
+    try {
+      console.log(`Scraping specials for: ${row.name}`);
+      await scrapeCategory(row.id, row.category_id);
+    } catch (error) {
+      console.error(`Failed to scrape specials for ${row.name}:`, error);
+    }
+  }
+  
+  console.log('✅ Specials scrape completed');
+}
+
+export async function startScheduler(): Promise<void> {
+  // Stop any existing jobs
+  stopScheduler();
+  
+  const settings = await getSettings();
+  
+  // Schedule regular store scrapes (every N hours)
+  const scrapeIntervalCron = `0 */${settings.scrapeIntervalHours} * * *`;
+  const storeJob = cron.schedule(scrapeIntervalCron, scrapeAllStores);
+  scheduledJobs.push(storeJob);
+  console.log(`📅 Scheduled store scraping: every ${settings.scrapeIntervalHours} hours`);
+  
+  // Schedule specials scraping (daily at specified time)
+  const [hours, minutes] = settings.scrapeSpecialsTime.split(':');
+  const specialsCron = `${minutes} ${hours} * * *`;
+  const specialsJob = cron.schedule(specialsCron, scrapeSpecials);
+  scheduledJobs.push(specialsJob);
+  console.log(`📅 Scheduled specials scraping: daily at ${settings.scrapeSpecialsTime}`);
+  
+  // Initial scrape on startup (after 10 seconds)
+  setTimeout(() => {
+    console.log('🚀 Running initial scrape...');
+    scrapeAllStores().catch(console.error);
+  }, 10000);
+}
+
+export function stopScheduler(): void {
+  scheduledJobs.forEach(job => job.stop());
+  scheduledJobs = [];
+  console.log('🛑 Scheduler stopped');
+}
+
+export async function restartScheduler(): Promise<void> {
+  console.log('🔄 Restarting scheduler...');
+  stopScheduler();
+  await startScheduler();
+}
+
+// Manual trigger functions for admin
+export async function triggerStoreScrape(storeId: number): Promise<void> {
+  console.log(`🔧 Manual scrape triggered for store ID: ${storeId}`);
+  await scrapeStore(storeId);
+}
+
+export async function triggerAllStoresScrape(): Promise<void> {
+  console.log('🔧 Manual scrape triggered for all stores');
+  await scrapeAllStores();
+}